Classification and Regression-based Machine Learning Approach to Predict Mine Water Quality Index

This work proposes a data mining-based prediction and development of the water quality index in mining areas. A mathematical equation for the index and predicted model is derived quantitatively in the study. Predicting water quality often involves applying conventional data mining techniques like classification and regression. Predictive learning and testing models can be evaluated using previous monitoring in real-time datasets and implementing k-fold cross-validation methods. The "decision trees" classification methodology outperforms other classification methods with 97.30% and 99.50% accuracy for training and testing model validation. MAE, RMSE, MSE, and R-squared are used in regression analysis to evaluate prediction accuracy and model performance. Regression model errors are absent with an R-squared value of 1. The present research showcases the efficacy of data mining techniques in accurately estimating mine water quality. These findings help improve mine water quality management.

View article page

Download PDF